The best part about this is that looking at the code, it is totally something that I can see myself do. A bit of Python plus some useful packages really goes a long way.
I really emplore you to find some open data and just set your own challenges. I did that with Citibike open data and that snowballed into a 300-line 17-hour script that won me a trip to California. I've never heard the term Marey Chart, even though I'm now doing my third course related to railway transportation. We just call them time-space diagrams or blocking diagrams if they include the train's block occupancy times, like this visualization I made in Matlab a month ago:
I messed with the TfL APIs a little bit, they're pretty good. But that was more of a software development exercise when I was job hunting: http://rjw-tfl-bikes.herokuapp.com/ (this just plots the locations of bicycle hire on a map). Fine, I'll have another look. Maybe the fact that I can share it on here will motivate me... I didn't know the name either, just seen them referred to with regard to Tufte.
I also had some fun with the TfL API, mostly to get my foot in the water regarding GET requests / REST services / XML. Try to find a big spreadsheet and try to prove/disptove a hypothesis, or visualize the data in a new way. Here, check this out. Let me know what you come up with! :)